We have talked about several types of photon interactions with matter.

Going from
lowest energy to highest energy we have classical scatter, which occurs at very low
energy, then we have the photoelectric effect, then as we go higher in energy we
have Compton Scatter, and now as we go even higher in energy we will talk about a
process called pair production.

1
In today’s lecture we are going to identify and describe the process of pair
production. In addition, we will review the implications of all of the various
interaction processes in imaging, radiation shielding, and radiation treatment. We
will summarize everything and at the very end we will look at one more process that
fits off to the side from the others.

2
Let’s start with a qualitative view of pair production. Based on our classification of
interaction processes, pair production is a Type 3a process. In pair production, a
photon interacts with the electric field that surrounds the atomic nucleus and is
completely absorbed.

3
In pair production we have a photon transforming into an electron and positron pair.
I remember when I took my undergraduate quantum mechanics course and the
instructor introduced pair production. This did not make sense. All of a sudden you
have something from nothing but as far as conservation of energy and conservation
of momentum is concerned - it’s allowable.

We have a photon coming in. The interaction takes place in the vicinity of a
nucleus in order to conserve momentum. This photon transforms itself into an
electron and positron pair. How much energy do we need on the part of the photon?
First of all, some of the photon energy gets transformed to matter. How much
energy gets transformed to matter? 1.022 MeV, which is the rest energy of the
positron plus the rest energy of the electron. So the first 1.022 MeV of the photon
goes into producing the positron and the electron. The remaining energy becomes
kinetic energy that’s shared by the electron and the positron.

4
Here’s a diagram of the process. A photon comes in with energy hν. We produce an
electron and we produce a positron. We take the photon energy and subtract off the
2m0c2, which corresponds to the mass of the positron and electron pair, and the
remaining energy is kinetic energy that’s shared between the electron and the
positron.

5
Some of the kinetic energy goes into the nucleus, but because the nucleus is so
much more massive than the electron or the positron, we can generally ignore the
kinetic energy transferred to the nucleus.

We need energy of 2m0c2 to create the pair, so we can identify a threshold energy
for pair production to take place. If the incident photon has energy of less than
1.022 MeV, we don’t have enough energy to produce the pair, so pair production
cannot occur for energies less than 1.022 MeV.

Because some of the momentum is transferred to the nucleus, and we are not sure
exactly how much, we really can’t determine the relative angle between the electron
and the positron. We just know an electron and positron are produced. The electron
and the positron interact with target material. These are secondary interactions. We
will talk about those charged particle interactions in some subsequent lectures.
These charged particles will produce more ionizations and deposit more energy.

6
In some respects, the process of pair production can be looked upon as an inverse
process to the production of Bremsstrahlung. Let’s recall the process of
Bremsstrahlung production. In the production of Bremsstrahlung, an electron
undergoes a transition between two positive energy states, emitting a photon. In
pair production the electron can be looked at as undergoing a transition from a
negative energy state, creating a positron, to a positive energy state, absorbing a
photon.

Mathematically, the theory behind Bremsstrahlung and the theory behind pair
production are closely related and are usually treated together.

7
How do we determine the cross section? This requires a quantum mechanical
treatment. The first study was done by Heitler and he basically looked at it as the
reverse of Bremsstrahlung production. Recall the analogy we presented earlier. In
Bremsstrahlung, we have an electron being deflected, producing photons, whereas
in pair production, we have a photon coming in, producing the electron and the
positron.

Heitler got this expression for the differential scattering cross section for pair
production: dκ/dΩ is Z2 over 137 (Recall that 137 is the fine structure constant from
quantum mechanics) multiplied by r02 (the classical radius of the electron squared)
divided by 2π multiplied by the square of the electron rest energy multiplied by an
atomic form factor for pair production.

8
The atomic form factor is a complicated function of momentum, energy, and angle
of both the positron and the electron. We are not really going to be giving you the
full-blown expression for that form factor. We will not worry about what that
expression is but we are going to be interested in the dependence of the form factor,
and hence dκ/dΩ, on angle. We are going to be interested in the dependence of
dκ/dΩ on energy and we will be interested in the dependence of dκ/dΩ on atomic
number Z.

9
Let us first look at the angular distribution of the ejected electron and positron. For
incident photon energies near the threshold, that is, 2m0c2, the angular distribution is
rather complicated, but as the photon energy increases, the positron and electron are
ejected in a more forward direction, as would be expected from conservation of
momentum.

10
Moving on to the energy distribution, we are unable to predict the precise energy
distribution unless we could determine the kinematics in detail.

But, on the average, each of the two ejected particles gets about half of the available
energy. A slight asymmetry in the energy distribution is observed because the
nucleus attracts the negatively-charged electron, but repels the positively-charged
positron. Thus a small amount more kinetic energy is given to the positron. The
difference in kinetic energies is typically less than 0.25 MeV.

11
Here’s a picture of what a typical energy distribution graph looks like.

The graph (taken from Johns & Cunningham) indicates the relative probability per
fractional energy interval. The relative probability is normalized so that the integral
under the curve is unity, since the total relative probability is 1.0.

12
Notice that the split in energies changes with energy.

For low energy photons, for example, 5 MeV photons, we see that the most probable
energy split is roughly 50/50 – half the kinetic energy goes to the positron and half
the kinetic energy goes to the electron. So if we have a 5 MeV photon, we subtract
off 1.022 MeV to produce the pair giving us a little under 4 MeV of kinetic energy
to be split between positron and electron. The most probable distribution is that 2
MeV goes to the electron and 2 MeV goes to the positron.

At higher energies the split is more asymmetric, about 25% to 75% being the most
probable. As you can see it’s a fairly broad peak, leaving a uniform distribution
really out to about 15% to 85% all the way across.

For 5 MeV photons the most probable energy distribution is peaked at 50/50. For
high energy photons the most probable energy distribution is roughly 25/75; but in
general, it’s flat from about 15% to 85%.

13
Notice that, not counting the small asymmetry we mentioned earlier, the energy
distribution is symmetric; either particle has the same relative probability so they’ll
share the energy equally. The average energy transferred to either particle is half of
the difference between the incident photon energy and the energy needed to produce
the pair. This enables us to calculate energy transfer cross sections. We start with
the average energy transferred, which, in this case, is hν-2m0c2. We divide that by
the energy of the incident photon hν and multiply by the linear attenuation
coefficient κ, and that gives us the energy transfer coefficient. Remember how we
calculate energy transfer coefficients. Take the average energy transferred, divide it
by the incident photon energy, and multiply it by the linear attenuation coefficient
for the interaction.

14
We can use this information to work out some problems. Here’s an example: Again,
we are going to try to track the energy. We want to know how many positrons in the
energy range 6.9 to 7.1 MeV are set in motion when we have a layer of carbon of
thickness 3 X 1026 atoms/m2 placed in a 20 MeV beam of 106 photons. Let’s see
how to solve this problem.

15
Our task is to calculate the number of pairs that are created. The number of pairs is
given by the number of incident photons, that is, the photon fluence multiplied by
the fraction of those photons per unit absorber thickness that undergo interactions
multiplied by the absorber thickness. And again, we have a thin absorber.

We could use the exponential attenuation form if the absorber were thick, but for the
time being, we are going to assume that the absorber is thin, and we will calculate
the number of pairs in this manner.

16
We know the photon fluence; we were given that it is 106 photons. We know what
the absorber thickness is, 3 x 1026 atoms per square meter. What is the attenuation
coefficient for pair production? We can look that up in Johns & Cunningham. That
number can be found in Table A-4b in the appendix of Johns & Cunningham. I
hope you are taking advantage of your Johns & Cunningham tables as they are very
useful. We find that the attenuation coefficient is 0.1321 x 10-28 meters squared per
atom.

So the number of pairs produced is the fluence, 106, multiplied by the attenuation
coefficient, 0.1321 x 10-28, times the absorber thickness, 3 x 1026 atoms per square
meter. Doing the multiplication gives us that 3.96 times 103 positron-electron pairs
are produced.

17
Next we need to find out how the energy is distributed between positrons and
electrons.

First of all, how much energy is available? The energy that’s available is 20 MeV,
the incident photon energy, minus the energy that’s needed to produce the positron-
electron pair, which is 1.02 MeV. So, approximately 19 MeV is available to be
given to the pair.

Next, we want to determine how many positrons of a given energy are produced.
We’re looking at positrons of energies between 6.9 and 7.1 MeV; let’s call that 7
MeV. 0.368 is the fraction of energy given to the positron.

18
What is the relative probability per energy interval of generating a positron with this
energy? According to this graph, for a fraction of energy equal to 0.368, we’re
looking at a relative probability per energy interval of about 1.09. The curve is
relatively flat here and somewhat insensitive to energy in this range.

That’s why we weren’t worried about how many decimal places we have in the
energy. Notice, it’s pretty flat and roughly the relative probability per energy
interval is about 1.09.

19
We have to multiply this probability times the fractional energy interval. The
fractional energy interval is 0.2 MeV out of 19 MeV, or 0.2 parts out of 19. So the
probability is 0.0115. Multiply this probability by the total number of positrons to
give us 45 positrons in this energy interval that are generated via pair production.

20
What happens after pair production takes place. The positrons and electrons
transfer energy to the target. We will see that described in the next few lectures.
Through the processes of ionization and excitation, the positron deposits energy and
eventually comes to rest. At that time it annihilates with an electron generating 2
photons of 0.511 MeV energy ejected at 1800 to one another. We have already seen
the applications of positrons to imaging, although the energy of a positron involved
in pair production may be much higher than those energies we see in imaging.

21
Another process that we need to be aware of is called triplet production. It occurs
when pair production occurs in the vicinity of an electron rather than a nucleus. In
triplet production a positron-electron pair is generated and also the electron is
ejected. The threshold energy for triplet production is 2.04 MeV, 1.022 MeV to
create the pair, and 1.022 MeV to conserve momentum and energy.

22
The attenuation coefficient for triplet production is approximately 1/6 that as for
pair production, but it is usually included in calculated attenuation coefficients for
pair production. So generally we are not going to worry so much about triplet
production when we look at these interactions. We are going to focus mainly on
pair production at the higher energies.

23
What can we say about the various dependences of the pair production attenuation
coefficient? This graph plots the pair production electronic attenuation coefficient
as a function of photon energy. Notice that it is on a log-log plot.

24
Two very important observations can be made. One is the threshold energy at 1.022
MeV. If the photon energy is less than the threshold 1.022 MeV, pair production
will simply not occur. Notice, however, that for energies above the threshold
relative increase in attenuation coefficient as a function of energy is relatively flat,
meaning that the attenuation coefficient is relatively independent of energy.

In summary, we have a threshold of 1.022 MeV, a dramatic increase in attenuation

coefficient just above the threshold, and rather flat energy dependence at higher
energies.

25
The other thing we need to look at is the Z dependence of the attenuation
coefficient. Pair production occurs in the force field of a nucleus, so a larger
nuclear mass means a higher probability of interaction. The nuclear mass is roughly
2Z, consequently, the mass attenuation coefficient is roughly proportional to Z.

Note that earlier in the lecture I said that the atomic attenuation coefficient is
proportional to Z2. In going from atomic attenuation coefficient to mass attenuation
coefficient, we divide by Z, so the mass attenuation coefficient is proportional to Z.

26
In summary, we can say the following: When pair production occurs, a photon
interacts in the vicinity of the nucleus creating an electron-positron pair. Pair
production has an energy threshold of 1.022 MeV required to produce the pair, with
the remaining energy shared by the electron and the positron. On the average each
of the two gets half of the energy.

27
Triplet production also occurs. But it is usually included in consideration with pair
production.

The mass attenuation coefficient for pair production is roughly proportional to Z.

The attenuation coefficient increases with energy immediately above the threshold
and then levels off at higher energies. Eventually, two annihilations of photons are
produced. Each photon has energy of 0.511 MeV. These annihilation photons can
cause ionization elsewhere at some distance from the location of the interaction.

28
Let us now tie everything together, take a step backward, and look at all the
attenuation coefficients.

The total attenuation coefficient is the sum of the individual attenuation coefficients
for each interaction, the attenuation coefficient for the photoelectric effect, plus the
attenuation coefficient for coherent scatter, plus the attenuation coefficient for
Compton scatter, plus the attenuation coefficient for pair production. In addition we
can write similar expressions for the mass attenuation coefficient, the electronic
attenuation coefficient, and the atomic attenuation coefficient. Moreover, there will
be similar expressions for energy transfer coefficients and energy absorption
coefficients.

Remember how we go from linear attenuation coefficient to energy transfer

coefficient to energy absorption coefficient. To obtain the energy transfer
coefficient we take the amount of energy transferred to kinetic energy of charged
particles, divide it by the energy of the incident photon, and multiply it by the
appropriate linear attenuation coefficient. Of course in each attenuation process the
energy transferred to kinetic energy of the charged particles is going to be a
different value depending on the particular process. To obtain the energy absorption
coefficient, we need to know what fraction of that energy transferred to charged
particles is absorbed versus what fraction is reradiated in the form of
Bremsstrahlung. So that’s how we’re going from one coefficient to another
coefficient.

29
This is really a neat graph. It summarizes all the attenuation coefficients for
oxygen.

The dotted line is Rayleigh scatter. Notice it tails off pretty quickly with energy and
at no point is it the predominant interaction coefficient.

Photoelectric effect starts off as the predominant interaction coefficient and drops
off very rapidly with energy. We know that the dependence on energy is roughly
energy to the minus 3rd power.

Compton scatter starts off low and then flattens off and then tails off with higher
energy; remember it’s roughly 2 orders of magnitude change in attenuation
coefficient for 5 orders of magnitude of energy.

Pair production starts off at the threshold energy of 1.02 MeV and increases
gradually with energy.

So here is the sum of all of them. On this same graph we see the total attenuation
coefficient for tin. Notice we have this little blip here for the absorption edge
because of the binding energy of tin electrons.

30
Let’s summarize. Rayleigh, or coherent, scatter never has the largest attenuation
coefficient. The attenuation coefficient for Rayleigh scatter is greater than that for
Compton scatter at low energies. The attenuation coefficient for Rayleigh scatter
becomes greater than that for the photoelectric effect near about 100 keV, but by
that time Compton scatter has taken over. Finally, when Rayleigh scatter occurs,
there’s no energy lost in the medium.

31
Photoelectric absorption is the dominant interaction at low photon energies and for
high-Z materials. The probability of photoelectric absorption falls off very rapidly
with increasing energy. This probability is inversely proportional to the E3 power.
This probability increases dramatically with increasing Z. In particular, it is
approximately proportional to Z3.

32
The probability of absorption increases very sharply when the energy of the photon
matches the binding energy of an orbital electron. These are demonstrated as
photoelectric absorption edges. Finally, photoelectric absorption is a very efficient
energy transfer process because all the photon energy with the exception of the
amount of energy required to overcome the binding energy of the photoelectron gets
transferred to the electron. We don’t lose any energy to scattered photons or
anything like that.

33
Compton scatter is the dominant interaction in soft tissue in the photon energy range
used in radiation therapy. There is very little change in absorption probability with
energy. We notice less than 2 orders of magnitude decrease in probability as we go
from 1 keV to 100 MeV, which is 5 orders of magnitude, with the maximum
attenuation occurring around 100 keV.

34
Compton scatter is not a good energy transfer mechanism at low energies.
Remember that Compton scatter approaches the classical limit where there is no
energy transfer. Compton scatter is a more efficient energy transfer mechanism at
higher energies.

The mass attenuation coefficient for Compton scatter is virtually independent of

atomic number and virtually independent of energy.

35
And finally we have pair production. Pair production exhibits a threshold energy at
1.022 MeV. The probability of attenuation increases rapidly at energies above the
threshold but the probability of attenuation levels off at higher energies. The energy
is shared by the electron/positron pair. Either can have any energy from 0 to the
available energy which is the energy of the incident photon minus 1.022 MeV. On
the average the electron and positron have the same energy.

36
When the positron comes to rest, two annihilation photons are created, each with
energy of 0.511 MeV and ejected at 180°; consequently, we can have additional
ionization going on at a distance from the original interaction site. Pair production
is an efficient energy transfer process because once you overcome that 1.022 MeV
to produce the electron/positron pair, all the kinetic energy goes into the pair of
charged particles.

37
Here’s a nice table of energy transfer and energy absorption that also illustrates what we have been saying about
energy transfer and energy absorption. Notice the percentage of interactions by each process and the fraction of
energy that’s transferred by each process.

We start with coherent scatter and we see that the fraction of interactions reaches a maximum at about 30 keV,
but even at that energy, it’s nowhere near being the predominant interaction. And, of course, there is no energy
transfer in coherent scatter.

We see here the energy dependence of the photoelectric effect, starting from its being the predominant effect at
low energies, and its probability going to zero above 200 keV. Notice that, in general, the fraction of energy
transferred via the photoelectric effect is greater than the fraction of interactions that take place via the
photoelectric effect. The reason why this occurs is that almost all the energy of the incident photon is
transferred to the photoelectron, whereas for other processes, the energy transfer is not as efficient.

We observe that from 30 keV to 30 MeV, Compton scatter is the predominant interaction. Those are two good
numbers to remember, 30 keV – 30 MeV, as the energy range in which Compton scatter is the predominant
interaction in soft tissue.

38
Pair production only starts to be a significant component here about 15 MeV and
becomes the predominant interaction past 30 MeV, and notice again, that the
fraction of energy transferred via pair production is greater than the fraction of
interactions occurring. Analogous to what happens in the photoelectric effect, when
pair production occurs, a fixed amount of energy goes into overcoming an energy
barrier, with the rest of the energy being transferred to kinetic energy of charged
particles.

We finally look at the fraction of energy lost to Bremsstrahlung and see that at low
to moderate energies, very little energy is lost to Bremsstrahlung; that is, most of the
energy is absorbed by the charged particles, but at high energies, some of the energy
is re-radiated as Bremsstrahlung.

We need to know all of this if we are going to try to derive energy transfer
coefficients or energy absorption coefficients, or if we want to go from fluence to
dose, which is energy absorbed, which is ultimately going to be our goal. We are
going to be revisiting a lot of this later in the course when we talk about cavity
theory.

39
Let us conclude this lecture by examining some of the implications of all of this
information.

First of all, let’s look at implications in imaging. In the energies used in diagnostic
x-ray, much of the beam is in the energy range where photoelectric absorption is the
predominant interaction. Therefore, contrast, which is the difference in absorption,
is caused by differences in atomic number as well as in density giving rise to high-
contrast images. In particular, the differentiation between bone and soft tissue,
which corresponds to a difference of about 50% in atomic number, is greatly
magnified by the fact that attenuation goes as the cube of the atomic number. This
is why imaging works so well at low energies. This differential absorption giving
rise to contract is so dependent on Z.

Contrast media such as barium have K-edges near the energies of the diagnostic x-
rays, giving rise to further enhancement of absorption.

40
In the x-ray energy region used in radiation therapy, much of the beam interacts by
Compton scatter, so when we image using photons at therapy energies, contrast is
caused only by differences in density. Consequently, the images have much lower
contrast. Those of you who are going to work with therapy megavoltage portal
images are going to see that the image quality of a megavoltage image is pretty bad.
One reason is that there is so little difference in contrast between bone and soft
tissue.

41
We also need to look at the radiation that reaches the image receptor when we
image using radiation that interacts with the target.

When photoelectric absorption occurs, the radiation that is emitted is an electron

with a very short range. Consequently, the only radiation that penetrates the patient
and reaches the image receptor consists of photons that have not been involved in
any sort of interaction. Because of this, we know exactly where that photon came
from, retracing its path directly back to the source.

42
For higher energy photons that interact via Compton scatter, we could eject a
photon, which can also reach a detector. The problem is that this radiation carries
no information about where it came from. Consequently, it adds noise to the image
that’s produced at the image receptor. When Compton interactions occur, not only
do we have poor contrast because of differences in absorption but we also have
more noise because of the presence of scattered radiation.

These are some implications of the different energy ranges and the different
interactions in imaging.

43
What about implications of these interactions in shielding? In the diagnostic energy
range, the attenuation is highly dependent on Z, remember attenuation goes as Z3.
What do we use for shielding material in this energy range? The most effective
shield in the diagnostic energy range is a material with high Z. In fact, we use lead,
which has the highest Z of all stable isotopes.

But in the therapy energy range, in which Compton scatter is the predominant
interaction, all materials have the same attenuation per unit mass. Another way of
saying this is that equal masses attenuate equal amounts. We do not get any
advantage with high-Z material. As a consequence, we normally use standard
construction materials for shielding. We shield our therapy rooms with several feet
of concrete, because it’s easier to use several feet of concrete that using several
inches of lead.

Equal masses shield equal amounts.

44
Finally, what are the implications of the various interactions in radiation therapy? If
we are trying to treat with radiation in the orthovoltage range, that is around 100-
250 kVp, a large fraction of the interactions are in the photoelectric regime.
Absorption of energy is highly dependent on Z. Consequently for a given amount
of radiation, bone, a higher-Z material, absorbs much more energy than an equal
mass of soft tissue. However, when we are in the megavoltage range of energies,
where Compton scatter is the predominant interaction, the differences in energy
absorption (remember energy per unit mass) among the various forms of tissue are
negligible. Equal masses absorb equal amounts. An equal mass of bone absorbs the
same amount of radiation as an equal mass of soft tissue. Consequently the dose
delivered to bone equals the dose delivered to soft tissue to a very good
approximation.

I hope you keep in mind a lot of the implications of these different interactions.
That’s all very important especially when we go into Med Phys 2 and Med Phys 3
and we learn about these applications to both imaging and therapy.

45